Using two stream architecture to implement a classic action recognition method on UCF101 dataset
855
stars
252
forks
source link
If I take 64 x-channel and 64 y-channel optical flow images as input to motion cnn, should I make the out channel num of conv1_custom bigger? #66
Open
ljmiao opened 5 years ago
self.conv1_custom = nn.Conv2d(channel, 64, kernel_size=7, stride=2, padding=3, bias=False)
here, input channel will be 128, Will 64 be too small?